A Comparative Study of Integral Dose: Pelvic Irradiation

Proceedings of the 52nd Annual ASTRO Meeting IV. All patients received concurrent chemotherapy. The median dose to the pelvis including inguinal nodes was 45 Gy. The primary site and involved nodes were boosted to a median dose of 57.6 (45-64.8 Gy) All acute and late toxicities were scored according to the Common Terminology Criteria for Adverse Events, version 3.0. Results: All patients completed planned course of treatment. The median follow-up was 15 months (range, 1-34). Acute grade $ 3 GI, dermatological and hematological toxicities were 7 (22.6%) , 7 (22.6%), and 12 (38.7%), respectively. Seventeen patients required a treatment break with a median duration of 2 days (1-20). Twenty-nine pts (94%) had complete clinical response, of those 19 (61%) were confirmed by biopsy. Fourteen patients had post-treatment PET/CT at a median of 9 weeks (2-27) with complete metabolic response (CMR) in 10 pts. of those patients with CMR 2 failed distantly and 1 developed in situ recurrence in anal canal and had local resection only). Four patients had partial metabolic response on PET/CT. Two of these had persistence of disease while other two had complete clinical response, all confirmed pathologically. of the two patients with persistence of disease, one had salvage surgery and the other had a peri-rectal node failure, but declined salvage surgery, and received further radiation and chemotherapy. The 2-year colostomy-free survival, overall survival, freedom from local failure, and freedom from distant failure were 96.6%, 100%, 86.8%, 90.1%, respectively. Conclusions: Our outcome analysis shows that concurrently chemotherapy with PET/CT-based IMRT for treatment of anal carcinoma is feasible with excellent response rates and reduced toxicities when compared to historical controls. The follow-up PET/ CT scan appears encouraging for response assessment, but should be interpreted with caution and confirmed with biopsy as our small series suggests high false positivity. Author Disclosure: R.S. de Andrade, None; G.S. Deffoe, None; S. Beriwal, None; D.E. Heron, None.

147

Dosimetric Parameters Are Predictive of Acute Gastrointestinal Toxicity in Patients Treated with Concurrent Chemotherapy and Intensity Modulated Radiation Therapy for Anal Carcinoma

S. G. DeFoe, T. S. Kehwar, S. Beriwal, S. Won, H. A. Jones, D. E. Heron University of Pittsburgh Cancer Institute, Pittsburgh, PA Purpose/Objective(s): To determine whether dosimetric parameters are predictive of acute gastrointestinal (GI) toxicity in patients treated with intensity-modulated radiotherapy and concurrent chemotherapy for anal carcinoma. Materials/Methods: Fifty-eight patients with anal cancer were treated with concurrent chemotherapy and intensity-modulated radiation therapy (IMRT) at the University of Pittsburgh Cancer Institute between March 2003 and March 2009. The median radiation dose to the gross tumor volume (GTV) and involved regional lymph nodes was 55.8 Gy, (range, 45-67.6). The small bowel was defined on the planning CT and included all opacified and non-opacified bowel loops within the pelvis. Patients were monitored weekly during treatment for acute GI toxicity, which was scored using the Common Terminology Criteria for Adverse Events, version 3.0. Regression models with multiple independent predictors were used to test associations of clinical factors and dosimetric parameters against clinically significant GI toxicity (grade $3). Factors significantly associated with grade $3 GI toxicity on logistic regression were fitted to a normal tissue complication probability (NTCP) model, using a logit function. Results: The median patient age was 57.5 years, (range, 50.0-68.2). Seventeen patients (29.3%) had stage T3-4 disease, 18 (31%) patients were node positive. Three patients were HIV positive. Two patients (3.4%) experienced no acute GI whereas, 20 (34.5%) experienced grade 1, 20 (34.5%) experienced grade 2, and 16 (27.6%) experienced grade 3, none experienced grade 4. On univariate analysis, none of the clinical factors (stage, age at diagnosis, type of chemotherapy, treatment interruption) were predictive of grade $3 GI toxicity. However, univariate analysis of volume of bowel receiving 10 Gy (V10), 15 Gy (V15), 20 Gy (V20), 30 Gy (V30) and 40 Gy (V40) found the volume of bowel receiving 30 Gy (V30) and 40 Gy (V40) both predicted for grade $3 GI toxicity, p = 0.037 and 0.044, respectively. Patients whose V30 was . 310 cc grade $3 GI toxicity was 38.9% compared to 9.1% if V30 remained #310 cc, p = 0.02. If V40 was #70 cc, acute grade $3 toxicity was 6.3% versus 35.7% if V40 .70 cc, p = 0.04. Conclusions: Dosimetric parameters such as V30 and V40 are predictors of grade $3 GI toxicity in anal cancer patients treated with concurrent chemotherapy and IMRT. Incorporation of these dose constraints into IMRT planning will help further reduce the toxicity of treatment. Author Disclosure: S.G. DeFoe, None; T.S. Kehwar, None; S. Beriwal, None; S. Won, None; H.A. Jones, None; D.E. Heron, None.

148

A Comparative Study of Integral Dose: Pelvic Irradiation

H. K. Warkentin, L. Capelle, G. C. Field, K. Joseph, B. Warkentin Cross Cancer Institute, Edmonton, AB, Canada Purpose/Objective(s): Although deterministic effects of low dose irradiation on normal tissues are not yet well understood, mathematical models and epidemiological data support an increased risk of secondary cancer induction with increased integral dose. Consideration of integral dose levels may thus be important in the clinical assessment of the risks and benefits of a particular treatment. The use of intensity modulated radiotherapy (IMRT) has led to a change in the volume of normal tissue receiving a low radiation dose. IMRT techniques allow delivery of high doses to the planning target volume (PTV) while reducing the doses to surrounding critical structures. However, to achieve this, a larger volume of normal tissue often receives a low dose. Studies have reported that the increased number of monitor units with IMRT increases integral dose due to additional linac head leakage and possibly neutron production. This study quantifies integral dose due to the dose distribution alone, thus complementing existing calculations of integral dose due to treatment delivery. Materials/Methods: Using the same dosimetric criteria, dynamic MLC IMRT, volumetric modulated arc therapy (VMAT), and helical tomotherapy (HT) plans were created for 18 patients, for a total of 54 treatment plans. The treated sites were anal canal, rectum, and prostate. Integral dose was calculated as the product of the mean dose and the volume for all tissue excluding the PTV. The integral dose from each treatment technique was analyzed using a repeated measurements (RM) ANOVA test and Tukey HSD test. For a smaller group of 10 anal canal patients, further analysis was performed to compare the three IMRT techniques to a conformal technique. Results: The mean integral doses for IMRT, VMAT, and HT are, respectively: 294 k Gy*cm3, 279 k Gy*cm3, 318 k Gy*cm3. The RM-ANOVA test results indicate that the difference in integral dose is strongly correlated with treatment technique (p \ 0.0001, F = 33.2). The Tukey HSD test indicates that the increased integral dose for HT compared to IMRT and VMAT is significant (p \ 0.01), and also that IMRT relative to VMAT is significant (p \ 0.05). When compared to a conformal technique, the RM-ANOVA

S69

I. J. Radiation Oncology d Biology d Physics

S70

Volume 78, Number 3, Supplement, 2010

also indicates a strong correlation with treatment technique (p\0.0001, F = 21.6). The Tukey HSD test indicates that the HT plans are not significantly different from the conformal plans. Conclusions: Our results for this sample indicate that shifting from conformal to IMRT techniques does not necessarily result in an increase in integral dose, and may in fact lead to a reduction. HT planning resulted in an 8% and 14% increase in mean integral dose compared to dynamic MLC IMRT and VMAT, respectively. If integral dose is of concern to the clinician, explicit care should be taken to minimize the dose to all normal tissue when using IMRT techniques. Author Disclosure: H.K. Warkentin, None; L. Capelle, None; G.C. Field, None; K. Joseph, None; B. Warkentin, None.

149

High Quality and Low Dose Cone-beam CT Imaging with Anisotropic Regularized FDK Reconstruction Algorithm

H. Lee1, R. Lee2, T. Suh3, W. Jeong3, L. Xing1 1

Stanford University, Stanford, CA, 2Ewha Women’s University, Seoul, Republic of Korea, 3The Catholic University of Korea, Seoul, Republic of Korea Purpose/Objective(s): In current radiation therapy, CBCT scans are repeatedly performed on the same patient during the whole treatment course, which lasts 4 to 6 weeks. While acquiring projection data with a lower mAs protocol is an effective way to reduce radiation dose incurred to the patients, excessive noise of projection data leads to significantly degraded image quality of CBCT when it is reconstructed by the well-known Feldkamp-Davis-Kress (FDK) algorithm. Therefore, we propose an improved FDK reconstruction algorithm with an anisotropic regularization to enhance image quality of CBCT using low dose projections. Materials/Methods: The algorithm firstly applied a modified filtering that integrates shepp and logan filter into window function on all projections for impulse noise removal. Anisotropic regularization method was then applied on the filtered projection images. Anisotropic diffusion filter (ADF) was used as anisotropic regularization and minimized by the standard steepest gradient decent optimization strategy. Minimizing ADF of the filtered projection image indicates that edges having high contrast relative to the surroundings are preserved and noisy pixels having low contrast are smoothed. The preserving parameter to adjust the separation between noise-free and noisy areas was adaptively determined by calculating the cumulative histogram of the gradient magnitude of the updated projection image in each steepest gradient decent step. In this work, the parameter value was set to be 90% of the cumulative histogram. With these anisotropic regularized projections, backprojection operation was finally performed. The performance of our anisotropic regularized FDK algorithm was evaluated with phantom and patient data sets acquired from a low-dose protocol (10mA/10ms) and compared with conventional FDK algorithm. Results: High quality CBCT images were obtained with an order of magnitude lower dose using the proposed algorithm. A quantitative analysis of the images with low dose shows that the contrast-to-noise ratio with spatial resolution comparable to images reconstructed with projections acquired from a high-dose protocol (80mA/12ms). Artifacts seen in traditional FDK algorithm were markedly suppressed in images reconstructed by the proposed algorithm. Conclusions: The proposed algorithm reduces radiation dose and provides high contrast-to-noise CBCT images by removing noisy corrupted areas and preserving the edges in the repeated CBCT scans. The work may have significant implication in image guided and adaptive radiation therapy as CBCT is used repeatedly. Author Disclosure: H. Lee, This work was supported by the National Research Foundation of Korea, NRF, (Grant no. 2008-1768-2 and Grant no. 2009-1311-1) funded by Ministry of Education, Science, and Technology., C. Other Research Support; R. Lee, None; T. Suh, None; W. Jeong, None; L. Xing, None.

150

Dosimetric Evaluation of a Volume Segmentation Algorithm for MRI-based Treatment Planning for Head and Neck Cancer

C. Chang, B. K. Teo, M. Altschuler, A. Lin, T. C. Zhu University of Pennsylvania, Philadelphia, PA Purpose/Objective(s): The objective of this study is to evaluate the dosimetric accuracy of IMRT dose calculation using only a T1 MR image. MRI based radiotherapy dose calculation techniques have previously been applied to the pelvic and brain regions. For head and neck region, tissue heterogeneity can have a more pronounced effect on dose calculation accuracy. Using an automated segmentation based algorithm with thresholding, a heterogeneous pseudo CT image is derived from a T1 MRI volume with four distinct typesair, lung, tissue, and bone. Each class is assigned a CT number corresponding to fixed relative electron densities of 0, 0.25, 1.0, and 2.0, respectively. We assess the accuracy of dose calculation using this four level electron density model to account for tissue heterogeneity. Materials/Methods: A fast 3D volume segmentation algorithm is developed to delineate the different tissue types from a MR image. In addition to the external body contour, the program provides a tool to obtain internal MR boundaries onto which the initial physiciandrawn tumor contours can be mapped and correlated. Direct comparison is made between the actual CT and MR derived pseudo CT to assess segmentation accuracy. Radiotherapy dose was first calculated and optimized on the CT using the Eclipse treatment planning system (Varian Medical) with heterogeneity correction and subsequently recalculated using the pseudo CT image for comparison. Results: For T1 MRI, we found a good one-to-one correspondence for air, lung, and tissue between CT and MRI numbers. For our MRI sequence, preliminary data show that this correspondence is: 0-15 for air; 15-60 for lung, and 60-500 for tissue. There is an overlap between lung and bone: 40-60 for bone and 15-60 for lung probably due to the similarly decreased hydrogen contents in these materials compared to tissue. This degeneracy can be removed through image processing by first identifying the lung contour and segmenting regions that are either inside or outside the lungs. Comparison of the coverage to head and neck IMRT treatment planning show good concordance of the dose volume histograms between CT and MR image calculated plans. The mean and maximum dose values of the structures were typically within less than 2% between the two methods. Conclusions: MRI-based treatment planning has proven to be feasible for sites where tissue heterogeneity is insignificant (e.g., prostate). Our study extends that to sites with substantial tissue heterogeneity. Using a multi-organ segmentation algorithm, we are able to convert the MR image into a four level CT with electron density information for dose calculation. Preliminary verification for head and neck patients show that MRI based dose calculation is a feasible method and can potentially be useful for future integrated MRI/Linac systems. Author Disclosure: C. Chang, None; B.K. Teo, None; M. Altschuler, None; A. Lin, None; T.C. Zhu, None.